A Study Of Driving Risk Based On Dangerous Goods Transportation Vehicles' Critical Incident Events

In recent years,China's social and economic development has been rapid,and the nation's transportation demand for dangerous goods such as oil and natural gas has also been expanding,which has brought great challenge to transportation safety.Although the number of road traffic accidents caused by dangerous goods transport vehicles has not been large,the number of casualties and property losses is serious,therefore,many studies at home and abroad define road dangerous goods transportation accidents as Low Probability High Consequence(LPHC)events.At present,the analysis of transportation accidents of road transport dangerous goods at home and abroad shows that more than 70% of accidents are caused by human factors,and driver skills have the greatest impact.Therefore,research on driving risk of dangerous goods transport vehicles is the current research hotspot.The research on the driving risk of traditional road transportation dangerous goods is mainly based on accident data for transportation accident analysis and path optimization.However,traffic accidents are small probability events during actual transportation,alternative methods are usually used to study traffic safety and driving risks.Such as critical incident events.At present,the research on driving risk at home and abroad mainly focuses on cars.There are few relevant literatures on natural driving experiments for dangerous transport vehicles,the accumulation of experimental data and characterization indicators are insufficient.Therefore,it is the main study on driving risk of dangerous goods transport vehicles based on the low-probability and high-risk accident risk characteristics of dangerous goods transport vehicles and the characterization indicators of traffic safety events.The research on the risk of dangerous goods transportation process,this paper comprehensively adopts natural driving data collection,questionnaire survey and other methods,based on statistical analysis,data mining and other technologies,establishes a driving risk model based on dangerous traffic vehicles driving safety events,and evaluates their effectiveness.The specific research is as follows:Firstly,based on the critical incident events,the premise of driving risk can be characterized,and the dynamics and accident characteristics of tank semi-trailers can be classified into three categories: rapid acceleration,rapid deceleration,sharp turn;for typical driving safety.The event characterization index(horizontal and vertical)collects the driving data of 20 drivers for one month according to the natural driving data platform,and determines the critical incident events characterization index and threshold range for the dynamic characteristics of the tank semi-trailer and the typical dangerous driving behavior theory model.To analyze the type and frequency of critical incident events,and provide a basis for the subsequent driving risk identification model.Secondly,through the collection of tank semi-trailer driver behavior questionnaire(DBQ),factor analysis is carried out to divide the active infringement and violation factors and driving ability errors and negligence factors representing the driving style,respectively,using the scores and factors of the two factors after rotation.The variance contribution rate is used to calculate the factor composite score of 20 drivers.The subjective driving risk of the sample driver is determined according to the factor comprehensive score,and the verification sample is provided for the training sample set of the subsequent driving risk identification model and the test sample set provides the verification reference.Finally,based on the naive Bayesian identification model,constructing a driving risk identification model based on driving safety events.Using Matlab's function toolbox,the frequency and risk level of critical incident events of the 15 days before the test are selected as the training sample set for training.The frequency and risk level of critical incident events 15 days after the test was verified as a test sample set,the accuracy of the model was verified by questionnaire survey results,and the classification accuracy rate reached 80.00%.The driving risk identification model based on dangerous goods transport vehicles' critical incident events proposed in this paper has theoretical guidance and practical application value for monitoring and early warning of dangerous behavior and improving driving safety.